Diabetic macular edema and age-related macular degeneration (AMD) are the two major causes of visual loss in developed countries. While laser therapy for these and other diseases has prevented loss of visual function in many individuals, disease progression and visual loss following suboptimal treatment is common. For AMD, there is unambiguous evidence that incomplete laser photocoagulation of the border of a choroidal neovascular lesion is associated with an increased risk for further visual loss, while treatment beyond the borders unnecessarily destroys viable, central photoreceptors, further degrading visual function.
As a concrete example, in eyes with juxtafoveal choroidal
neovascularization (CNV) secondary to ocular histoplasmosis, only
of eyes with laser treatment that covered the foveal side of the
lesion with a narrow 
treatment border suffered severe
visual acuity loss, while 
of eyes with either some of the
foveal side untreated, or a wide border of treatment on the foveal
side suffered severe visual loss (MPS, 1995). Similar results have
been reported for AMD. Building on the recommendations proposed in
Macular Photocoagulation Studies, clinicians generally attempt to
correlate angiographic data with biomicroscopic images using crude,
time-consuming, potentially error-prone methods. In a recent practical
review (Neely, 1996), the author suggests that ``... to assist you
in treatment (of neovascular AMD), project an early frame of the
fluorescein angiogram onto a viewing screen. Use the retinal vessels
overlying the CNV lesion as landmarks. I suggest tracing an image of
the CNV lesion and overlying vessels onto a sheet of onion skin
paper. It takes a little extra time, but I find it helps to clarify
the treatment area.'' Accordingly, precise identification of the
treatment border during laser therapy by correlating the
biomicroscopic image with fluorescein angiographic data (where the
lesion extent is better delineated, see for example Figure
1) should be beneficial for maximizing post-treatment
visual function. Diagnosis and treatment relies on synthesizing
clinical data derived from fundus biomicroscopy with angiographic
data, but methods for correlating these data, and for direct guidance
of laser therapy, are not well-developed.
We are exploring techniques to overlay angiographic data on the real-time biomicroscopic slitlamp fundus image in order to guide treatment for eye disease (for example, to better define and visualize the edges of choroidal neovascular membrane, and improve identification of focal areas of leakage in diabetic macular edema). The biomicroscopic fundus image will be ``augmented'' in real time with available angiographic data. Text display and a ``virtual pointer'' will be incorporated into the augmented reality display to facilitate teaching, telemedicine, and real-time measurement and image analysis. Moreover, image superposition will allow for direct comparison with previous images to judge disease progression (for example, to judge progression or stability of AMD or cytomegalovirus retinitis---a common, blinding disease afflicting patients with acquired immunodeficiency syndrome) and allow for real-time identification of prior treatment areas. This technology is straightforwardly extended to an indirect-ophthalmoscope-based or operating-microscope-based system to facilitate correlative, teaching, and telemedicine applications in these environments.
In this report, we describe considerations for the design and implementation of an ophthalmic augmented reality environment and report on our preliminary studies in model systems.
Figure: Monochromatic photographic
(left) and fluorescein angiographic (right) image of an eye with
age-related macular degeneration. The optic nerve is at the far left
of each photograph, with the fovea located centrally. Note that the
angiogram conveys additional information regarding areas of leaky
blood vessels.